Assessing the validity of driver gene identification tools for targeted genome sequencing data
Felipe Rojas-Rodriguez1, Marjanka K Schmidt1,2,3, Sander Canisius1,4
1Division of Molecular Pathology, The Netherlands Cancer Institute-Antoni van Leeuwenhoek Hospital, 1066 CX Amsterdam, The Netherlands.
Bioinformatics Advances
|May 29, 2024
Summary
Seven cancer driver gene identification tools were tested on targeted sequencing data. Four tools (OncodriveFML, OncodriveCLUSTL, 20/20+, dNdSCv, ActiveDriver) are recommended for targeted sequencing, while two are not suitable.
Area of Science:
- Genomics
- Cancer Genomics
- Bioinformatics
Background:
- Whole-exome sequencing (WES) is standard for cancer driver gene identification.
- Targeted sequencing offers cost-effective depth for large cancer studies.
- The applicability of WES-based tools to targeted sequencing data is unknown.
Purpose of the Study:
- Evaluate the performance of existing driver gene identification tools on targeted sequencing data.
- Determine the suitability of popular tools for analyzing targeted cancer sequencing panels.
- Provide recommendations for optimal tool selection in targeted sequencing studies.
Main Methods:
- Assessed seven popular driver gene identification tools.
- Utilized whole-exome sequencing data from 14 cancer types (TCGA).
- Created targeted datasets mimicking MSK-IMPACT and B-CAST panels.
Main Results:
- Tool performance varied significantly on targeted sequencing data.
- Differences in background mutation rate modeling impacted tool validity.
- OncodriveFML, OncodriveCLUSTL, 20/20+, dNdSCv, and ActiveDriver showed reliable performance.
- MutSigCV and DriverML were found unsuitable for targeted sequencing.
Conclusions:
- Driver gene identification tools developed for WES may not perform well on targeted sequencing data.
- Specific tools are recommended for accurate driver gene discovery in targeted sequencing.
- Understanding background mutation rate modeling is crucial for tool selection.
